« Previous | Contents | Next »
Listen
PUBLIC ATTITUDES TO THE ENVIRONMENT IN SCOTLAND - TECHNICAL REPORT
Chapter 5 - CODING AND DATA PROCESSING
5.1 CODING
The EAS questionnaires comprised largely pre-coded questions. Of the 66 questions on questionnaire A and 65 questions on questionnaire B, only 3 of the questions were fully open. However, many others required answers that did not fit into the suggested pre-code categories to be listed and coded.
Most of the coding was straightforward. Of the three open questions, one (asking respondents what they understood sustainable development to be about) produced wide ranging answers. Although a summary code frame was developed for this question, it was decided also that a full listing of all responses would be helpful to see to what extent individuals provided or linked different combinations of ideas to sustainable development. This was provided to the Scottish Executive as a separate document.
There was a slight concern arising from the summarised answers to this question that some of the responses may have been affected by previous questions on the questionnaire. In particular there were relatively high levels of mention of the idea that sustainable development encompassed the replacement of trees when cut down. These questions had followed on immediately from questions about Forest Stewardship and Forest Certification (answered by just 11% of the sample). However, the full transcript of answers to the sustainable development question indicated that the terms being used bore a close resemblance to the definitions often seen attached to cards and other paper products, whereby consumers are informed that '2 trees will be planted for every 1 cut down.' As such, the responses were not simply regurgitations from the earlier questions.
For other questions, the listing was undertaken by our Coding Manager who then developed a draft set of code frames. These were inspected by the project manager at the Scottish Executive and, based on a preliminary analysis of around 2,000 completed and data entered questionnaires, some summary codes were agreed.
These summary codes were used in the preliminary analysis prepared by George Street Research for the Scottish Executive. However, the full dataset includes the original more detailed codes and these can be referred to if needed when conducting any detailed analysis.
5.2 DATA PREPARATION
All data was keyed in using the SNAP survey analysis package. Each variable had range checks to ensure that data outside a valid range could not be entered and there was an automatic routing facility to ensure that the correct routing for the questionnaires was followed.
A 20% verification of the punched data was undertaken. Once the data had been input, a preliminary analysis was run which was used as the basis for checking what additional checks were needed on the data. The list of edit and coding checks undertaken was provided to the project manager. A copy of this document is provided as
appendix 5. The key point to mention relates to the 12 instances where the respondents age was missing from the dataset. As this was used in the weighting of the data, it was agreed that the age would be imputed, bearing in mind other information known about the household. The ID numbers where this occurred have been provided to the project manager.
The data files were provided to the project manager as a SPSS sav file.
A separate weighting file was provided in excel format. The following chapter of this report indicates what information was included in the weighting file and how the weights were devised. Both household and individual weights were provided. Careful consideration needs to be given to which weight is used in relation to each question. Essentially the study was undertaken as a sample of individuals. However, clearly some of the data collected relates to the behaviour of the household and so for some questions it might be more appropriate to use the household weighting variable. For example, many of the waste and recycling questions in questionnaire B are about what may be regarded as household behaviours.
In addition, a second excel file was provided which gave the Mosaic code and full address postcode for every household where an interview was completed. This can be used for more advanced data interpretation and modelling.
The links between all three of these files are through two fields contained on each dataset:
1 field, consisting of a four digit number referred to as the point number. The point numbers ranged from 1001 to 6070. The first digit here relates to the rural/urban indicator and ranges from 1 - 6. The following three digits relate to the unique sample point number with a range from 1 - 800.
1 field, consisting of a 2 digit number referred to as the ID or SIS number. These range from 01 to 10.
These two fields have been combined in the dataset in a unique identifier called v1.
« Previous | Contents | Next »